Closed Adel-Moumen closed 7 months ago
I think we should not have speechbrain/benchmarks
that depends on speechbrain main
branch. This is too hard for us to maintain. By having a git speechbrain submodule that uses a specific SB version (in this case develop), we can update the version of the submodule and update this repo accordingly. We can reduce the amount of work to do since we will update as long as we are updating speechbrain develop branch and not when we are releasing a new version.
I proceed with the merge (according to what we said in yesterday meeting) so that I can move on with MP3S PR @mravanelli.
This PR adds SpeechBrain as a submodule.
This way, we have a direct reference to Speechbrain inside of the repo. It is easier for us developer, and for the end user to use the same exact SpeechBrain version.
We can easily reflect in benchmarks the new changes in SpeechBrain develop by updating the hash commits.